August 12, 2026
Artificial intelligence is rapidly becoming a tool used by many in daily life. From grocery shopping lists, to work out plans and meeting note summaries—the possibilities of what AI can create for its users seem endless. The legal profession has followed suit. Lawyers are using AI to quickly summarize a factual scenario, clean up an email, or find a case exactly on point.
Despite its clear benefits, AI also has significant drawbacks. For example, large language models (LLMs) run the risk of producing hallucinations. Hallucinations are false or inaccurate outputs generated by AI, produced with confidence in its accuracy yet rooted in gaps in the system’s training data.
Consequently, hallucinations have proven particularly dangerous for attorneys. Time and again legal news sources have reported on fake cases cited in legal filings. In other situations, AI has generated a real case but fabricated the information or quotation attributed to the case. Either the court has discovered the error independently, or the attorney has admitted to using AI for research or drafting. Judges have not hesitated to issue sanctions against attorneys who cite cases that don’t exist or when an attorney cites or quotes a case, yet the holding or quotation appears nowhere in the cited material.
Sometimes citing a hallucinated case or inaccurate quotation or premise results in a slap on the wrist or a statement about the misconduct from a judge in open court. But the consequences can be much worse. Federal judges throughout the country have issued monetary sanctions ranging anywhere from $1,000 to $10,000 against attorneys who filed documents that they used AI to write. Judges have also issued non-monetary sanctions against attorneys and the firms they work for, such as requiring them to implement AI policies and submit proof of the same to the court. While monetary sanctions are certainly steep, the harms of inappropriate AI use reach further. Attorneys could face reputational harm within their firm, with judges they practice before, and in the legal community at large. Some attorneys have, voluntarily or at the suggestion of their employer, resigned after misusing AI.
The above examples are a cautionary tale about what can happen when AI is misused, but this is not to say attorneys should never use it. Rather, attorneys who use AI should do so responsibly. For legal research specifically, consider AI tools embedded in established platforms like LexisNexis or Westlaw. And always double check the cases generated by AI–first, confirm the case exists, and second, if it does, read it to ensure it stands for the proposition AI claims or contains the quote AI pulled. Many law firms and judges allow the use of AI so long as it is used responsibly and the manner of use is disclosed openly. Many law firms have already implemented mandatory AI training, both on its uses and on the ethical issues that can arise and have been implemented in firm platforms that maintain confidentiality like Harvey. In January of this year, the ABA came out with a checklist of AI best practices for law firms. The basic tenets of AI use are simple: use the right tools, ensure security and confidentiality, review and check for accuracy and quality, and require final approval by a human lawyer. AI is a great tool–but it is only as great as the user makes it.
Jessica Conroy (LAW ’26) is a Student Editor for the Temple 10-Q.